An Extended Analysis of Chinese Translation Students’ Attitudes towards Translation Technologies through the Technology Acceptance Model (TAM): Exploring the Influence of Demographic Variables on TAM Constructs
1. Introduction📘
1.1 Translation Technologies (TT)✨
As technology increasingly permeates every aspect of society, the synergy between human translation and technology has become a defining feature of modern translation practices. Advancements in translation technologies (TT) have fundamentally reshaped the tools and methodologies available to the translation industry. These technologies, which vary in their levels of automation and application, can broadly be categorized into two primary branches: Machine Translation (MT) and Computer-Assisted/Aided Translation (CAT) (see, e.g., Alcina (2008), p. 80; Doherty (2016), p. 947). They have revolutionized how translators manage tasks, input and output information, and transfer language units. According to Doherty (2016) (p. 950), this technological evolution has resulted in “unprecedented gains in terms of increased translator productivity and consistency, greater global language coverage, and greater support for improving international communication and distribution.” Given these transformative effects, the integration of TT into educational settings is not merely beneficial but essential for equipping students with the skills needed in the modern translation industry.
1.2 Technology Acceptance Model (TAM)✨
To explore the factors influencing the adoption of technology in education and other fields, researchers frequently turn to the Technology Acceptance Model (TAM). Introduced by Fred Davis over 25 years ago, TAM has emerged as a widely used framework for understanding technology acceptance (Marangunić & Granić (2014), p. 81). TAM posits that users’ motivation to adopt technology is influenced by three primary factors: Perceived Ease of Use (PEU), Perceived Usefulness (PU), and Attitude toward Technology Using (ATT). According to Davis (1989) (p. 320), PU is defined as “the degree to which a person believes that using a particular system would enhance their job performance” and PEU as “the degree to which a person believes that using a particular system would be free of effort.” These factors are interrelated, with PEU and PU both shaping the overall user attitude toward technology. In this context, PEU is also believed to have a direct influence on PU (Marangunić & Granić (2014), p. 85).
These constructs form the theoretical backbone of TAM, which has been widely employed to study technology adoption across a variety of disciplines and professions. For example, Scherer et al. (2019) synthesized findings from 114 empirical studies involving 34,357 teachers to examine TAM’s applicability in digital education. Their meta-analysis highlighted the significance of TAM’s core variables (PEU, PU, and ATT) in predicting behavioral intentions (BI), namely, a person’s subjective probability of performing a particular behavior (Fishbein & Ajzen (1975), p. 288) and technology use, demonstrating the model’s robust explanatory power. Beyond education, TAM has also been validated in domains such as healthcare, business, and public services, further underscoring its versatility.
1.3 Mounting Applications of TAM in Translation Studies (TS)✨
In the context of translation studies (TS), the exploration of factors that influence student translators’ intention to use MT, CAT, and other TT under the framework of TAM has gained increasing scholarly interest (see, e.g., Salloum et al. (2024), p. 527; Yang & Wang (2019), p. 116; Li et al. (2024), p. 2; Dianati et al. (2022), p. 14). Salloum et al. (2024) (pp. 536–537) integrated motivation and experience as external variables, highlighting their roles in shaping BI and PEU. Their findings revealed that increased experience with ChatGPT enhances perceptions of its utility and that motivation influences PEU uniformly across user groups. Yang & Wang (2019) (pp. 122–123) found PU to be a stronger predictor of BI than PEU, though an unsupported relationship between PEU and PU suggested mixed trust in machine translation (MT). Dianati et al. (2022) (pp. 25–26) identified instructors’ attitudes toward translation and interpreting (T&I) technologies as influenced by perceptions of usefulness, ease of use, and resources. However, challenges like cost, steep learning curves, and limited geographic scope underscored the need for broader institutional support and methodological refinement. Collectively, these studies validate TAM’s robustness but highlight areas for improvement, including more diverse samples, longitudinal designs, and qualitative methods.
1.4 Focus on Li et al. (2024)✨
Building on the foundation above, Li et al. (2024) examined the validation of TAM through a new perspective. Their study focused on 370 juniors and seniors majoring in Translation and Interpretation at public universities in China, a group selected for their frequent engagement with translation technologies as part of their academic curriculum, particularly during their third and fourth years. Data were collected via an online questionnaire, utilizing a convenience sampling technique appropriate for exploratory research with easily accessible populations. The sample size of 370 participants aligns with the recommended guidelines for structural equation modeling, ensuring sufficient statistical power for the analysis.
By investigating the factors influencing college students’ acceptance of translation technologies (TT), it is confirmed that PU, PEU, and ATT all play significant roles in shaping students’ acceptance of TT, particularly computer-assisted translation (CAT), suggesting that the results align with the foundational aspects of the Technology Acceptance Model (TAM). The study also extends TAM by identifying two additional factors: computer self-efficacy (CSE), referring to an individual’s belief in their ability to use computers effectively, and perceived enjoyment (PE), the extent to which using technology is enjoyable in its own right, as key determinants of students’ intention to use translation technologies (Li et al. (2024), pp. 4–5).
However, the study provided limited exploration of how demographic variables (such as age, gender, grade, and family residence) might have specific influence on students’ CSE, PE, PEU, PU, ATT, and BI in terms of translation technologies. Additionally, the extent to which these demographic factors might shape students’ overall perceptions and attitudes of TT remains unclear. Given the growing emphasis on personalized learning and the increasing role of technology in education, understanding how demographic factors impact students’ perceptions of translation technologies seems both relevant and important. Therefore, this project seeks to extend the efforts and fill the limitations in Li et al. (2024) by analyzing data from its questionnaire (publicly available at [data]) to answer the following questions:
RQ1. To what extent do demographic factors (age, gender, grade, family residence) influence students’ perceptions of computer self-efficacy (CSE) and perceived enjoyment (PE) in using translation technologies?
RQ2. To what extent do demographic variables (age, gender, grade, family residence) impact students‘ perceptions of perceived usefulness (PU) and perceived ease of use (PEU) of translation technologies?
RQ3. To what extent do demographic variables (age, gender, grade, family residence) impact students’ attitudes towards use (ATT) and behavioral intentions (BI) in the context of translation technologies?
The analysis will leverage data encompassing demographic variables (age, gender, grade, family residence) and TAM core constructs (PU, PEU, CSE, PE, ATT and BI). Using R Studio for data analysis and visualization, this project aims to enrich TAM’s theoretical framework within translation studies while offering practical insights into tailoring technology adoption strategies to diverse student populations.
1.5 Research Hypotheses✨
H1: Impact of Demographic Factors on Computer Self-Efficacy (CSE) and Perceived Enjoyment (PE)
Age
H1a: Age will have a significant impact on students‘ perceptions of computer self-efficacy, with older students perceiving themselves to have higher computer self-efficacy than younger students.
H1b: Age will have a significant impact on students’ perceived enjoyment of using translation technologies, with older students reporting lower perceived enjoyment than younger students.
Gender
H1c: Gender will significantly influence students‘ perceptions of computer self-efficacy, with male students perceiving higher self-efficacy in using computers than female students.
H1d: Gender will significantly influence students’ perceived enjoyment of using translation technologies, with male students reporting higher enjoyment than female students.
Grade Level
H1e: Students‘ grade level will have a significant effect on their computer self-efficacy, with higher-grade students having greater confidence in their computer skills.
H1f: Students’ grade level will have a significant effect on their perceived enjoyment of translation technologies, with higher-grade students experiencing lower enjoyment compared to lower-grade students.
Family Residence
H1g: Family residence (whether rural or urban) will have a significant impact on computer self-efficacy, with students from urban areas having higher computer self-efficacy than students from rural areas.
H1h: Family residence will influence students‘ perceived enjoyment of translation technologies, with students from urban areas reporting greater enjoyment than those from rural areas.
H2: Impact of Demographic Factors on Perceived Usefulness (PU) and Perceived Ease of Use (PEU)
Age
H2a: Age will significantly influence students’ attitudes toward the perceived usefulness of translation technologies, with older students perceiving translation technologies as more useful than younger students.
H2b: Age will significantly influence students‘ attitudes toward the perceived ease of use of translation technologies, with younger students perceiving translation technologies to be easier to use than older students.
Gender
H2c: Gender will significantly impact students’ attitudes toward the perceived usefulness of translation technologies, with male students viewing translation technologies as more useful than female students.
H2d: Gender will significantly impact students‘ attitudes toward the perceived ease of use of translation technologies, with male students perceiving translation technologies as easier to use than female students.
Grade Level
H2e: Grade level will significantly impact students‘ attitudes toward the perceived usefulness of translation technologies, with higher-grade students perceiving translation technologies as more useful.
H2f: Grade level will significantly impact students‘ attitudes toward the perceived ease of use of translation technologies, with lower-grade students perceiving translation technologies to be easier to use than higher-grade students.
Family Residence
H2g: Family residence will significantly influence students‘ attitudes toward the perceived usefulness of translation technologies, with students from urban areas perceiving translation technologies as more useful than those from rural areas.
H2h: Family residence will significantly influence students‘ attitudes toward the perceived ease of use of translation technologies, with students from urban areas perceiving translation technologies as easier to use than those from rural areas.
H3: Influence of Demographic Factors on Attitudes towards Use (ATT) and Behavioral Intentions (BI)
Age
H3a: Age will significantly influence students‘ attitudes toward translation technologies, with older students exhibiting more positive attitudes than younger students.
H3b: Age will significantly influence students‘ behavioral intentions to use translation technologies, with older students having stronger behavioral intentions than younger students.
Gender
H3c: Gender will significantly impact students‘ attitudes toward translation technologies, with male students showing more positive attitudes than female students.
H3d: Gender will significantly influence students‘ behavioral intentions to use translation technologies, with male students exhibiting stronger behavioral intentions than female students.
Grade Level
H3e: Grade level will significantly impact students‘ attitudes toward translation technologies, with higher-grade students showing more positive attitudes than lower-grade students.
H3f: Grade level will significantly influence students‘ behavioral intentions to use translation technologies, with higher-grade students having stronger behavioral intentions than lower-grade students.
Family Residence
H3g: Family residence will significantly influence students‘ attitudes toward translation technologies, with students from urban areas showing more positive attitudes than those from rural areas.
H3h: Family residence will significantly influence students‘ behavioral intentions to use translation technologies, with students from urban areas having stronger behavioral intentions than those from rural areas.
2. Data and Methods📕
2.1 Variables in the Dataset✍️
This study includes two types of variables: Demographic Variables and Key Constructs based on the Technology Acceptance Model (TAM). The detailed descriptions are as follows:
Demographic Variables:
[Grade]: Students’ educational grade.
Categories:
Junior = 1
Senior = 2
[Gender]: Gender of the student.
Categories:
Male = 1
Female = 2
[Age]: Age of the student.
Categories:
Below 18 years = 1
18–22 years = 2
Above 22 years = 3
[Family Residence]: Family’s place of residence.
Categories:
Urban = 1
Rural = 2
Key Constructs (TAM Variables):
Key constructs are based on Likert-scale questionnaire items linked to the Technology Acceptance Model (TAM). The constructs and their corresponding measurement items are as follows:
For each TAM construct, aggregated mean values are computed by averaging the responses from their corresponding Likert-scale items. These aggregate variables are used for analysis:
[CSE_M]: Mean value of items CSE1–CSE4.
[PEU_M]: Mean value of items PEU1–PEU3.
[PE_M]: Mean value of items PE1–PE3.
[PU_M]: Mean value of items PU1–PU3.
[ATT_M]: Mean value of items ATT1–ATT4.
[BI_M]: Mean value of items BI1–BI3.
2.2 Methodology✍️
Step 1: Data Preprocessing and Hypotheses Visualization🍀
Load Dataset
Import the selected dataset into R using relevant functions read_excel()
Code
# Import the datasetlibrary(readxl)data_raw <-read_excel("data_raw.xlsx")# View the dataset to confirm successful importhead(data_raw)
# Load required librarylibrary(dplyr)# View the summary of the dataset to identify NA valuessummary(data_raw)
Grade Gender Age Family
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:1.000 1st Qu.:2.000 1st Qu.:2.000 1st Qu.:1.000
Median :2.000 Median :2.000 Median :2.000 Median :2.000
Mean :1.573 Mean :1.854 Mean :2.046 Mean :1.676
3rd Qu.:2.000 3rd Qu.:2.000 3rd Qu.:2.000 3rd Qu.:2.000
Max. :2.000 Max. :2.000 Max. :3.000 Max. :2.000
CSE1 CSE2 CSE3 CSE4
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000
Median :3.000 Median :3.000 Median :3.000 Median :3.000
Mean :3.446 Mean :3.281 Mean :3.143 Mean :3.305
3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.000 Max. :5.000 Max. :5.000
PEU1 PEU2 PEU3 PE1 PE2
Min. :1.000 Min. :1.00 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.00 1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000
Median :4.000 Median :4.00 Median :4.000 Median :4.000 Median :4.000
Mean :3.697 Mean :3.67 Mean :3.589 Mean :3.524 Mean :3.492
3rd Qu.:4.000 3rd Qu.:4.00 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.00 Max. :5.000 Max. :5.000 Max. :5.000
PE3 PU1 PU2 PU3
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000
Median :4.000 Median :4.000 Median :4.000 Median :4.000
Mean :3.489 Mean :3.781 Mean :3.797 Mean :3.835
3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.000 Max. :5.000 Max. :5.000
ATT1 ATT2 ATT3 ATT4
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000
Median :4.000 Median :4.000 Median :4.000 Median :4.000
Mean :3.724 Mean :3.751 Mean :3.643 Mean :3.735
3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.000 Max. :5.000 Max. :5.000
BI1 BI2 BI3 CSE_M
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.000
Median :4.000 Median :4.000 Median :4.000 Median :3.250
Mean :3.743 Mean :3.735 Mean :3.776 Mean :3.294
3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.000 Max. :5.000 Max. :5.000
PEU_M PE_M PU_M ATT_M
Min. :1.000 Min. :1.000 Min. :1.000 Min. :1.000
1st Qu.:3.000 1st Qu.:3.000 1st Qu.:3.333 1st Qu.:3.250
Median :3.833 Median :3.667 Median :4.000 Median :4.000
Mean :3.652 Mean :3.502 Mean :3.805 Mean :3.714
3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000 3rd Qu.:4.000
Max. :5.000 Max. :5.000 Max. :5.000 Max. :5.000
BI_M
Min. :1.000
1st Qu.:3.333
Median :4.000
Mean :3.751
3rd Qu.:4.000
Max. :5.000
Code
# Check for missing values in each columnmissing_values <-sapply(data_raw, function(x) sum(is.na(x)))print(missing_values)
# Check for the percentage of missing values in each columnmissing_percentage <-sapply(data_raw, function(x) mean(is.na(x)) *100)print(missing_percentage)
Before diving into the statistical analysis, it’s crucial to establish a clear conceptual framework for the research by mapping the expected relationships between demographic variables (age, gender, grade and family residence) and Tam constructs (CSE, PE, PU, PEU, ATT, and BI). The graph visually illustrates how demographic factors are hypothesized to influence students’ perceptions and attitudes towards translation technologies, providing a solid foundation for the analysis that follows.
Code
# Install and load igraphif (!require(igraph)) install.packages("igraph")library(igraph)# Define nodes (demographic variables and TAM constructs)nodes <-data.frame(name =c("Age", "Gender", "Grade", "Residence", "CSE", "PE", "PU", "PEU", "ATT", "BI"))# Define edges based on hypotheses (relations between demographic factors and TAM constructs)edges <-data.frame(from =c("Age", "Age", "Gender", "Gender", "Grade", "Grade", "Residence", "Residence", "Age", "Age", "Gender", "Gender", "Grade", "Grade", "Residence", "Residence", "Age", "Age", "Gender", "Gender", "Grade", "Grade", "Residence", "Residence"),to =c("CSE", "PE", "CSE", "PE", "CSE", "PE", "CSE", "PE", "PU", "PEU", "PU", "PEU", "PU", "PEU", "PU", "PEU", "ATT", "BI", "ATT", "BI", "ATT", "BI", "ATT", "BI"),label =c("H1a", "H1b", "H1c", "H1d", "H1e", "H1f", "H1g", "H1h", "H2a", "H2b", "H2c", "H2d", "H2e", "H2f", "H2g", "H2h", "H3a", "H3b", "H3c", "H3d", "H3e", "H3f", "H3g", "H3h"))# Create the graph objectg <-graph_from_data_frame(d = edges, vertices = nodes, directed =TRUE)# Define node colors based on the variable typeV(g)$color <-ifelse(V(g)$name %in%c("Age", "Gender", "Grade", "Residence"), "lightcoral", "lightblue")# Adjust node sizes and label sizesV(g)$size <-ifelse(V(g)$color =="lightcoral", 35, 30) # Larger nodes for lightcoral group (Age, Gender, Grade, Residence)V(g)$label.cex <-0.6# Smaller label size for better readabilityV(g)$label.color <-"black"# Node label color# Adjust edge arrow size and label sizeE(g)$arrow.size <-0.6# Slightly larger arrows for better visibilityE(g)$label.cex <-0.5# Smaller edge labels for clarityE(g)$label.color <-"darkgreen"# Edge label color# Layout: Make the graph more spacious by adjusting the edge lengthslayout <-layout_with_fr(g) # Fruchterman-Reingold layout for spacing# Adjust layout to ensure more space between the nodeslayout <- layout *1.2# Increase the spacing between the nodes# Save the graph as a customized PNG imagepng("hypotheses_graph.png", width =1600, height =1200, res =300)plot(g, layout = layout, vertex.frame.color ="gray70", # Subtle frame for nodesmain ="Hypotheses Visualization", # Add the title heremargin =0.1) # Add space around the graphdev.off() # Close the plot device and save the image
Purpose: Examine how demographic variables influence TAM constructs by exploring patterns, trends, and potential relationships between demographic variables and TAM constructs.
Format data by creating a new dataset that reshapes the data for analysis.
Code
# Load necessary librarieslibrary(ggplot2)library(dplyr)# Load the datasetdata_dot_plots <- readxl::read_excel("data_raw.xlsx")# Select only the relevant columns: 4 demographic variables and aggregated TAM constructsdata_dot_plot <- data_raw %>%select( Grade, # Column for educational grade Gender, # Column for gender Age, # Column for age Family, # Column for family residence CSE_M, # Aggregated TAM construct: CSE mean PEU_M, # Aggregated TAM construct: PEU mean PE_M, # Aggregated TAM construct: PE mean PU_M, # Aggregated TAM construct: PU mean ATT_M, # Aggregated TAM construct: ATT mean BI_M # Aggregated TAM construct: BI mean )# Save the processed dataset to confirm the changeswrite.csv(data_dot_plot, "data_dot_plot.csv", row.names =FALSE)# Check the structure of the new datasetstr(data_dot_plot)
Use ggplot2 to draw dot plots in an effort to visualize how each demographic variable affects the TAM constructs.
Code
# Load necessary librarieslibrary(ggplot2)library(dplyr)library(patchwork)# Load the processed datasetdata_dot_plot <-read.csv("data_dot_plot.csv")# Define a function to create dot plots for a given TAM construct and demographic variablecreate_dot_plot <-function(data, tam_construct, demographic_var, y_labels, title_suffix) {# Define the number of unique categories in the demographic variable n_categories <-length(unique(data[[demographic_var]]))# Generate a color palette with the correct number of colors color_palette <-c("#1f77b4", "#ff7f0e", "#2ca02c") # Use three colors for 3 categories# Dynamically create breaks based on unique categories in the demographic variable unique_categories <-unique(data[[demographic_var]])ggplot(data, aes(x =!!sym(tam_construct), y =!!sym(demographic_var), color =as.factor(!!sym(demographic_var)))) +geom_jitter(width =0.2, height =0.1, size =1.5, alpha =0.7) +scale_color_manual(values = color_palette[1:n_categories], labels = y_labels) +# Apply the color palettescale_y_continuous(breaks = unique_categories, # Set the breaks dynamically based on the unique categorieslabels = y_labels # Corresponding labels for each unique category ) +labs(title =paste(tam_construct, "Vs", title_suffix), x = tam_construct,y = title_suffix,color = title_suffix ) +theme_minimal(base_size =12) +theme(legend.position ="top",plot.title =element_text(hjust =0.5) )}# Define demographic variables and labels for Grade, Gender, Age, and Family Residencedemographic_vars <-list(Grade =list(var ="Grade", labels =c("Junior", "Senior")),Gender =list(var ="Gender", labels =c("Male", "Female")),Age =list(var ="Age", labels =c("< 18", "18–22", "> 22")),Family =list(var ="Family", labels =c("Urban", "Rural")))# List of TAM constructs to generate plots fortam_constructs <-c("CSE_M", "PEU_M", "PE_M", "PU_M", "ATT_M", "BI_M")# Create the plots for each TAM construct and each demographic variableplots_grade <-lapply(tam_constructs, function(tam) {create_dot_plot(data_dot_plot, tam, demographic_vars$Grade$var, demographic_vars$Grade$labels, "Grade")})plots_gender <-lapply(tam_constructs, function(tam) {create_dot_plot(data_dot_plot, tam, demographic_vars$Gender$var, demographic_vars$Gender$labels, "Gender")})plots_age <-lapply(tam_constructs, function(tam) {create_dot_plot(data_dot_plot, tam, demographic_vars$Age$var, demographic_vars$Age$labels, "Age")})plots_family <-lapply(tam_constructs, function(tam) {create_dot_plot(data_dot_plot, tam, demographic_vars$Family$var, demographic_vars$Family$labels, "Family")})# Arrange and display the plots for Grade, Gender, Age, and Residence in grid layouts with 3 columnsfinal_grade_plot <-wrap_plots(plots_grade, ncol =3)final_gender_plot <-wrap_plots(plots_gender, ncol =3)final_age_plot <-wrap_plots(plots_age, ncol =3)final_family_plot <-wrap_plots(plots_family, ncol =3)# Display the four sets of plotsfinal_grade_plot
Code
final_gender_plot
Code
final_age_plot
Code
final_family_plot
Code
# Save plots as PNG filessave_plots <-function(grade_plot, gender_plot, age_plot, family_plot, output_dir =getwd()) {ggsave(file.path(output_dir, "dot_plot_grade.png"), grade_plot, width =12, height =10)ggsave(file.path(output_dir, "dot_plot_gender.png"), gender_plot, width =12, height =10)ggsave(file.path(output_dir, "dot_plot_age.png"), age_plot, width =12, height =10)ggsave(file.path(output_dir, "dot_plot_family.png"), family_plot, width =12, height =10)}# Save the plots to the working directorysave_plots(final_grade_plot, final_gender_plot, final_age_plot, final_family_plot)
Use dplyr and tibble to calculate the mean of 6 TAM constructs (CSE_M, PEU_M, PU_M, AT_M, SN_M, BI_M) for different demographic variables (Grade, Gender, Age, Family). The results will be summarized and presented in separate tables for each demographic variable.
Further visualize the relationship between demographic variables and TAM constructs to complement the dot plots.
Code
# Load the necessary librarieslibrary(ggplot2)library(tidyr)library(dplyr)# Convert the summary tables to long formatsummary_grade_long <- summary_grade %>%pivot_longer(cols = CSE_M:BI_M, names_to ="TAM_Construct", values_to ="Mean_Score")summary_gender_long <- summary_gender %>%pivot_longer(cols = CSE_M:BI_M, names_to ="TAM_Construct", values_to ="Mean_Score")summary_age_long <- summary_age %>%pivot_longer(cols = CSE_M:BI_M, names_to ="TAM_Construct", values_to ="Mean_Score")summary_family_long <- summary_family %>%pivot_longer(cols = CSE_M:BI_M, names_to ="TAM_Construct", values_to ="Mean_Score")# Create stacked bar plots with good comparabilitycreate_stacked_bar_plot <-function(data, x_var, x_label, title) {ggplot(data, aes_string(x = x_var, y ="Mean_Score", fill ="TAM_Construct")) +geom_bar(stat ="identity", color ="black", position ="stack") +# Add black bordergeom_text(aes(label =round(Mean_Score, 2)), # Add data labelsposition =position_stack(vjust =0.5), size =3) +scale_y_continuous(expand =c(0, 0)) +# Adjust y-axis range dynamicallylabs(title = title,x = x_label,y ="Mean Score",fill ="TAM Construct" ) +theme_minimal() +theme(axis.title =element_text(size =12, face ="bold"),axis.text =element_text(size =10),plot.title =element_text(size =14, face ="bold", hjust =0.5),legend.position ="bottom" )}# Create plots for each demographic variablegrade_plot <-create_stacked_bar_plot(summary_grade_long, "factor(Grade)", "Grade", "Stacked Bar Plot of TAM Constructs by Grade")gender_plot <-create_stacked_bar_plot(summary_gender_long, "factor(Gender)", "Gender", "Stacked Bar Plot of TAM Constructs by Gender")age_plot <-create_stacked_bar_plot(summary_age_long, "factor(Age)", "Age", "Stacked Bar Plot of TAM Constructs by Age")family_plot <-create_stacked_bar_plot(summary_family_long, "factor(Family)", "Family", "Stacked Bar Plot of TAM Constructs by Family")# Display the plotsprint(grade_plot)
Code
print(gender_plot)
Code
print(age_plot)
Code
print(family_plot)
Code
# Line Plot for Gradeggplot(summary_grade_long, aes(x =factor(Grade), y = Mean_Score, color = TAM_Construct, group = TAM_Construct)) +geom_line(linewidth =1) +# Replaced size with linewidthgeom_point(size =3) +labs(title ="Line Plot of TAM Constructs by Grade",x ="Grade",y ="Mean Score",color ="TAM Construct" ) +theme_minimal()
Code
# Line Plot for Genderggplot(summary_gender_long, aes(x =factor(Gender), y = Mean_Score, color = TAM_Construct, group = TAM_Construct)) +geom_line(linewidth =1) +# Replaced size with linewidthgeom_point(size =3) +labs(title ="Line Plot of TAM Constructs by Gender",x ="Gender",y ="Mean Score",color ="TAM Construct" ) +theme_minimal()
Code
# Line Plot for Ageggplot(summary_age_long, aes(x =factor(Age), y = Mean_Score, color = TAM_Construct, group = TAM_Construct)) +geom_line(linewidth =1) +# Replaced size with linewidthgeom_point(size =3) +labs(title ="Line Plot of TAM Constructs by Age",x ="Age",y ="Mean Score",color ="TAM Construct" ) +theme_minimal()
Code
# Line Plot for Familyggplot(summary_family_long, aes(x =factor(Family), y = Mean_Score, color = TAM_Construct, group = TAM_Construct)) +geom_line(linewidth =1) +# Replaced size with linewidthgeom_point(size =3) +labs(title ="Line Plot of TAM Constructs by Family",x ="Family",y ="Mean Score",color ="TAM Construct" ) +theme_minimal()
Step 3: Correlation Analysis🍀
Purpose: Quantify the strength and direction of relationships between demographic variables and TAM constructs.
Introduction to the Pearson Correlation Coefficient
The Pearson correlation coefficient (also known as Pearson’s r) is a measure of the linear relationship between two continuous variables. It quantifies the degree to which two variables move together.
Range: The value of the Pearson correlation coefficient ranges from -1 to +1.
+1: Perfect positive linear relationship (as one variable increases, the other increases in exact proportion).
-1: Perfect negative linear relationship (as one variable increases, the other decreases in exact proportion).
0: No linear relationship between the variables.
Perform the Correlation Analysis
Code
# Load necessary librarieslibrary(dplyr)# Set the seed for reproducibilityset.seed(123)# Round all numeric columns to 2 decimal placesdata_rounded <- data_dot_plot %>%mutate(across(everything(), ~round(. , 2)))# Perform Pearson Correlation Analysiscorrelation_matrix <-cor(data_rounded, method ="pearson")# Print the correlation matrixprint(correlation_matrix)
# Save the file to the working directorywrite.csv(correlation_matrix, file ="correlation_matrix.csv", row.names =TRUE)
Visualize the Correlation Analysis
Code
# Use the corrplot package for better visualizationif (!require(corrplot)) install.packages("corrplot")library(corrplot)# Create a correlation plot for better visualizationcorrplot(correlation_matrix, method ="circle", type ="upper", order ="hclust", col =colorRampPalette(c("red", "white", "blue"))(200))
Step 4: Visualization of Processed Data Relationships🍀
Purpose: Compare the observed relationships from the analyses with the initial hypotheses visualization by recreating a graph with adjusted edge sizes based on effect sizes (in this case, correlation values).
Based on the correlation results, create a graph using igraph where:
Node sizes represent variable significance
Edge weights represent the strength of relationships
Visualize the weighted nodes and edges
Code
# Load the Librarieslibrary(igraph)# Use the same nodes and edges as before, but adjust the edge weights# Double-check that the correlation matrix has the correct row and column namesrownames(correlation_matrix) <-c("Age", "Gender", "Grade", "Residence", "CSE", "PE", "PU", "PEU", "ATT", "BI")colnames(correlation_matrix) <-c("Age", "Gender", "Grade", "Residence", "CSE", "PE", "PU", "PEU", "ATT", "BI")# Now create a vector of weights for each edge using the correlation matrixedges$weight <-mapply(function(from, to) {# Get the correlation value between 'from' and 'to' variablesreturn(abs(correlation_matrix[from, to])) # Use absolute value of correlation}, edges$from, edges$to)# Create the weighted graph with the edge weightsg_weighted <-graph_from_data_frame(d = edges, vertices = nodes, directed =TRUE)# Check the edge weightsE(g_weighted)$weight
# Adjust node sizes and colors based on previous logic (optional)V(g_weighted)$color <-ifelse(V(g_weighted)$name %in%c("Age", "Gender", "Grade", "Residence"), "lightcoral", "lightblue")V(g_weighted)$size <-ifelse(V(g_weighted)$color =="lightcoral", 35, 30)V(g_weighted)$label.cex <-0.6V(g_weighted)$label.color <-"black"# Adjust edge arrow sizes and labels (optional)E(g_weighted)$arrow.size <-0.6E(g_weighted)$label.cex <-0.5E(g_weighted)$label.color <-"darkgreen"# Set edge widths based on the correlation values (scaled)E(g_weighted)$width <-E(g_weighted)$weight *5# Scale this as per your preference# Layout: Keep the same or use a different onelayout <-layout_with_fr(g_weighted)layout <- layout *1.2# Adjust spacing if needed# Save the weighted graph as a PNG imagepng("weighted_hypotheses_graph.png", width =1600, height =1200, res =300)plot(g_weighted, layout = layout, vertex.frame.color ="gray70",main ="Weighted Hypotheses Visualization",margin =0.1)dev.off() # Close the plot device and save the image
In this section, an overview of the key insights gathered from the data visualizations, including the dot plots, stacked bar plot, and line plot will be provided.
3.1.1 The Interpretation of the Dot Plots for TAM Constructs Vs Demographic Variables🔎
Note: CSE = Computer Self-Efficacy, PE = Perceived Enjoyment, PU = Perceived Usefulness, PEU = Perceived Ease of Use, ATT = Attitudes towards Use, BI = Behavioral Intentions.
In terms of Grade, both junior and senior students predominantly choose scores in the middle-to-upper range (mostly 3–4) across all six constructs: CSE, PEU, PE, PU, ATT, and BI. This consistency indicates a uniformity in performance between junior and senior students, with most respondents holding similar perceptions of these constructs. Interestingly, very few senior students rate any of the constructs as low as score 1, and none in junior students. This suggests that students, regardless of grade, tend to avoid the lower end of the scale, implying that they generally have neutral or positive views about these constructs, but negative perceptions (represented by a score of 1) are almost nonexistent. It is particularly surprising given the broad range of constructs measured and it would seem to contradict the hypotheses H1e and H1f, which assume that higher-grade students (seniors) should have higher confidence (in CSE) and lower enjoyment (in PEU and BI).
Overall, despite these differences, there is minimal variation in performance between junior and senior students. Both groups show similar tendencies toward middle-range scores, indicating a general consensus in how the constructs are perceived across grade levels. The lack of significant differences in how students rate the constructs points to a consistent trend in student attitudes, with only slight differences in the spread of scores for certain constructs.
Looking at Age, the results mirror those observed for Grade. Since grade and age are closely linked, it’s not surprising that the patterns are almost identical. Younger students in the junior grades and older students in the senior grades show similar perceptions of the constructs. This reinforces the idea that age doesn’t seem to have a strong independent effect on how students perceive these constructs; rather, it is more likely the result of the students’ academic progression.
When considering Gender, a few notable differences emerge. Female students tend to rate PU, ATT, and BI in a more favorable light compared to male students, with scores frequently above 4. This suggests that female students generally have a more positive attitude and stronger intentions regarding the use of technology. The reason is probably that female students feel more confident or motivated to engage with technology.
Finally, when examining Family Residence, there is only a slight difference between urban and rural students in their perceptions of the six constructs. Both groups exhibit similar patterns, with most responses concentrated around scores 3 to 4, indicating a general alignment in their overall perceptions. However, the distributions for CSE and PE in relation to family residence show more variability, particularly among rural students. For CSE, rural students display a broader spread of scores, ranging from very low ratings to high ones, suggesting that their perceptions of computer self-efficacy are more divergent. Some students appear to have a low level of confidence in their computer skills, while others feel quite competent. A similar trend is observed for PE, where rural students show a wider range of responses, possibly reflecting differences in access to technology or exposure to relevant learning experiences. These variations may point to disparities in resources or educational opportunities between urban and rural areas, which could influence how students perceive their own abilities or the ease of using technology.
In summary, the dot plots reveal that most students, regardless of grade or age, have similar middle-to-upper-range perceptions of the six constructs, with few selecting low scores. However, gender differences stand out, particularly with female students showing more positive attitudes toward PU, ATT, and BI, while male students exhibit lower ratings for CSE, PE, and ATT. Family residence shows only slight differences, with rural students displaying a more spread-out perception of CSE compared to their urban counterparts. These findings highlight the complex interplay of student demographics in shaping their perceptions across different constructs.
Interestingly, the results seem to diverge from some of the hypotheses. For instance, the expected influence of age on CSE and PE (H1a, H1b), which predicted that older students would report higher computer self-efficacy and lower enjoyment, does not fully materialize in the data. Additionally, while gender differences in CSE (H1c) and PE (H1d) were anticipated, the overall perceptions of male and female students do not show the magnitude of difference expected. Also, while the expectation was that older students would perceive translation technologies as more useful (H2a) and easier to use (H2b), and would have more positive attitudes (H3a) and stronger behavioral intentions (H3b), the data did not show significant differences based on age. Similarly, despite the hypotheses that male students would rate translation technologies as more useful and easier to use than female students (H2c, H2d), the overall perceptions of both genders were relatively aligned, with female students showing more positive attitudes toward PU, ATT, and BI, and no clear differences in ease of use. These discrepancies suggest that other factors, beyond demographics, may be influencing students’ perceptions, and further investigation is needed to better understand these patterns.
3.1.2 The Interpretation of the Stacked Bar Plots and Line Plots for TAM Constructs by Demographic Variables🔎
The Stacked Bar and Line Plots provide a complementary view of the data, with both visualizations revealing consistent patterns across the six TAM constructs. Given that both plots reveal similar trends, I have combined their interpretations to present a more comprehensive analysis of how demographic variables such as grade, gender, age, and family residence influence students’ perceptions of CSE, PEU, PE, PU, ATT, and BI.
Note: CSE = Computer Self-Efficacy, PE = Perceived Enjoyment, PU = Perceived Usefulness, PEU = Perceived Ease of Use, ATT = Attitudes towards Use, BI = Behavioral Intentions.
Note: 1= Junior, 2 = Senior.
In terms of Grade level, senior students, across all six constructs, consistently score higher than junior students.This trend is evident in both the stacked bar plots and line plots, where junior students tend to rate the constructs more favorably. Potential explanations may lie in that the relationship between academic experience and perception of technology might not always be linear. While senior students typically have more experience with technology, it’s important to recognize that increasing exposure over time does not always equate to more positive perceptions or higher willingness to engage with technology. Instead, factors such as overload, changing priorities, increased complexity, and the shift in confidence might contribute to the lower scores among senior students compared to their junior counterparts. This finding could certainly merit further investigation, as it challenges common assumptions about the relationship between academic experience and technology attitudes.
Note: 1= Male, 2 = Female.
When breaking down the data by Gender, the results become more nuanced. For CSE, PE, and PEU, male students score significantly higher than female students, indicating that they perceive themselves as more capable in computer skills and find translation technologies more enjoyable and easy to use. However, female students outscore male students in ATT and PU, suggesting that female students generally have a more positive attitude toward technology use and perceive it to be more useful. For BI (behavioral intention), both male and female students show similar levels of intention, with no significant difference between the two groups.
Note: 1= <18, 2 = 18-22, 3 = >22.
The influence of Age on the TAM constructs shows a general trend where students older than 22 years score higher across all six constructs compared to students aged 18 to 22 years. This suggests that older students tend to have more confidence in their computer skills, find translation technologies easier to use, and perceive them as more enjoyable and useful. However, the picture changes for students below 18. For this group, the data shows markedly different results, with younger students (below 18) generally rating the constructs much more negatively. This could be indicative of differences in experience or exposure to technology among younger students compared to their older counterparts.
Note: 1= Urban, 2 = Rural.
The Family Residence variable reveals contrasting trends across different constructs. Students from urban areas tend to score higher than those from rural areas on CSE, PE, PU, and BI, suggesting that urban students perceive themselves to be more confident in using technology and have stronger behavioral intentions to use translation technologies. However, rural students score higher in PEU and ATT. It is clear that the fact that access to the internet and opportunities to engage with advanced digital tools do not have a definite influence on their perceptions of usefulness or self-efficacy, despite their more favorable attitudes and ease of use.
In conclusion, when comparing the findings to the initial research hypotheses, several patterns align, while others diverge. Grade level somewhat supports H1e, as male students generally score higher across all constructs than female students, indicating greater computer self-efficacy. However, this does not align with H2c and H2d, as female students score higher than males on perceived usefulness and attitude. In terms of age, the results are mostly in line with H2a, with older students (above 22) scoring higher across constructs, suggesting they perceive translation technologies as more useful. However, the under-18 group scored lower across most constructs, contradicting H2a and H3a, challenging age as a straightforward predictor. Lastly, family residence supports H2g, showing that urban students score higher in CSE, PE, PU, and BI. However, rural students scored higher in ease of use and attitude, diverging from H2h, which predicted urban students would rate these higher. Therefore, the interplay between demographic variables and TAM constructs is more complex, with two or more variables potentially influencing students’ perceptions and behaviors simultaneously.
3.2 Correlation Analysis👀
Code
# Print the correlation matrixprint(correlation_matrix)
Age Gender Grade Residence CSE
Age 1.00000000 0.045508445 -0.16597314 0.043848799 -0.12277411
Gender 0.04550845 1.000000000 -0.05240029 -0.008397549 -0.09567503
Grade -0.16597314 -0.052400295 1.00000000 0.091415652 0.05933980
Residence 0.04384880 -0.008397549 0.09141565 1.000000000 -0.05043394
CSE -0.12277411 -0.095675026 0.05933980 -0.050433936 1.00000000
PE -0.05303877 -0.028922168 0.07375416 0.020321187 0.44520380
PU -0.06940855 -0.040639899 0.07377282 -0.058068631 0.52100748
PEU -0.03440580 0.070002580 0.03607100 -0.044300019 0.37488924
ATT -0.05414493 0.042740894 0.06011358 0.001110690 0.42988127
BI -0.05137559 0.003054452 0.05537809 -0.025797625 0.42172698
PE PU PEU ATT BI
Age -0.05303877 -0.06940855 -0.03440580 -0.05414493 -0.051375585
Gender -0.02892217 -0.04063990 0.07000258 0.04274089 0.003054452
Grade 0.07375416 0.07377282 0.03607100 0.06011358 0.055378091
Residence 0.02032119 -0.05806863 -0.04430002 0.00111069 -0.025797625
CSE 0.44520380 0.52100748 0.37488924 0.42988127 0.421726977
PE 1.00000000 0.58751543 0.69150980 0.76550291 0.611377333
PU 0.58751543 1.00000000 0.61203403 0.63897816 0.546281755
PEU 0.69150980 0.61203403 1.00000000 0.71556614 0.575799388
ATT 0.76550291 0.63897816 0.71556614 1.00000000 0.734316663
BI 0.61137733 0.54628175 0.57579939 0.73431666 1.000000000
Note: CSE = Computer Self-Efficacy, PE = Perceived Enjoyment, PU = Perceived Usefulness, PEU = Perceived Ease of Use, ATT = Attitudes towards Use, BI = Behavioral Intentions.
The correlation analysis reveals several key relationships between the variables in the dataset, providing insights into the factors that influence behaviors, perceptions, and performance. The correlation coefficients ranging from -1 to 1 are interpreted using the following thresholds:
0.1 to 0.3 (or -0.1 to -0.3) represents a weak correlation.
0.3 to 0.5 (or -0.3 to -0.5) represents a moderate correlation.
0.5 to 0.7 (or -0.5 to -0.7) represents a strong correlation.
Above 0.7 (or below -0.7) indicates a very strong correlation
Firstly, it confirms that PU and PEU are two foundamental pillars that influence translation students’ ATT and BI. This aligns with the core constructs of the TAM proposed by previous scholars. Specifically, the strong correlation coefficients between PU and ATT (0.72), PU and BI (0.61), PEU and ATT (0.77), and PEU and BI (0.61) illustrate the significant role that these two constructs play in shaping students’ attitudes toward technology adoption and their intentions to use it. These findings reinforce the theoretical validity of TAM in the context of translation studies.
Secondly, the findings of Li et al. (2024) are validated, as CSE and PE exhibit moderate correlations with ATT and BI. For instance, CSE shows a moderate positive relationship with ATT (0.43) and BI (0.42), indicating that students’ confidence in their technological abilities moderately influences their attitudes and intentions. Similarly, PE demonstrates moderate correlations with ATT (0.64) and BI (0.55), highlighting the role of the sense of fulfillment in driving students to accept and use translation technologies. These results validate the hypothesis that CSE and PE are important but secondary factors in influencing TAM constructs compared to PU and PEU.
Comparatively, this analysis highlights the weak connections between the four demographic variables (Grade, Gender, Age, and Family) and the six TAM constructs (PU, PEU, ATT, BI, CSE, and PE), which is a central focus of this study. While the correlations between demographic variables and TAM constructs are noticeably weaker than those observed among the TAM constructs themselves, Grade shows a weak negative correlation with CSE (-0.12). This suggests that students’ academic performance exerts a minimal influence on their confidence in using technology, implying that higher or lower academic achievement does not significantly shape students’ self-efficacy in navigating technological systems. Similarly, Age exhibits a negligible positive correlation with CSE (0.06), further affirming that age-related differences have little impact on students’ technological self-confidence.
Additionally, it is found that Gender and Family are only weakly correlated with the TAM constructs, reinforcing the limited role of demographic factors in this regard. For instance, the correlation between Family and BI is -0.03, while the correlation between Gender and PEU is -0.03, both of which are essentially negligible. Taken together, these results reveal that demographic factors such as age, gender, family background, and grade exert minimal influence on translation students’ attitudes toward technology and their behavioral intention to adopt it.
This weak influence of demographic variables is particularly noteworthy, as it shifts the spotlight onto the extended TAM constructs themselves—PU, PEU, CSE, and PE—which are shown to have a far greater impact on determining ATT and BI. While the strong and moderate correlations among the TAM constructs have been well-documented in prior research, this study uniquely highlights the marginal effect of demographic characteristics on students’ perceptions and behaviors regarding technology adoption.
3.3 Comparison with Initial Research Hypotheses👀
Note: CSE = Computer Self-Efficacy, PE = Perceived Enjoyment, PU = Perceived Usefulness, PEU = Perceived Ease of Use, ATT = Attitudes towards Use, BI = Behavioral Intentions.
In this section, we delve into the comparison between the weighted and unweighted hypotheses visualizations, focusing on how the relationships among demographic variables and the TAM constructs shift when edge sizes are adjusted based on effect sizes (correlation values). The weighted graph, with its varying edge lengths, allows us to better discern the strength of these connections, revealing which relationships hold the most significance. By contrast, the unweighted visualization treats all relationships equally, offering a more general, less detailed view. Through this comparison, we can pinpoint key changes and gain deeper insights into how demographic factors influence technology perceptions and adoption.
Starting with Residence, the weighted graph shows that it has the shortest edges to PU, PE, and PEU, indicating a closer relationship with these constructs. However, the fact that the node sizes for all variables are nearly the same and the edge and arrow sizes show little difference suggests that, the overall strength or impact of family residence on the various constructs might not be very distinct, at least in terms of the overall graph layout. This could mean that although Family Residence is positioned closer to certain constructs (like PU, PE, and PEU), the statistical correlation values may not differ significantly enough to create visually pronounced differences in edge thickness across the entire graph.
For Gender, the weighted visualization reveals a slightly more pronounced influence, especially with PEU, CSE, and ATT. The edges connecting Gender to PU and ATT are relatively shorter compared to the other constructs, reflecting stronger but still modest effect sizes. This suggests that gender-related factors, such as socialization, interests, or differing experiences with technology, might have a noticeable influence on how individuals perceive and interact with technology, particularly in terms of enjoyment and self-efficacy. However, even though the edges to these constructs are closer, the correlation values are not large enough to make these connections appear overwhelmingly strong in the graph.
Next, Grade and Age are positioned closely in the weighted graph, suggesting that they have a similar influence on the TAM constructs. This visual proximity implies that both factors have comparable impacts on how individuals perceive and engage with technology. Grade and Age likely reflect an individual’s stage in life or educational level, both of which could influence familiarity with and attitudes toward technology. However, the correlation matrix tells a more complex story: Grade shows positive correlations with all six TAM constructs, while Age exhibits negative correlations with the same constructs. Specifically, the coefficients from Grade to CSE, PEU, PE, PU, ATT, and BI are all negative, ranging from -0.12 to -0.05, whereas the coefficients for Age show positive relationships across the board, ranging from 0.04 to 0.07. This discrepancy highlights that while Grade and Age appear similar in the weighted graph, their actual statistical relationships with the TAM constructs diverge considerably.
In summary, the weighted hypotheses visualization does not offer a perfect one-to-one match with the numeric coefficients, but it successfully reflects the relative strength of the relationships. The overall trend from the correlation values is represented in the visual proximity of nodes and the thickness of edges: Family Residence shows weak and largely insignificant effects, while Gender has a slightly stronger influence. Grade and Age, although visually close, demonstrate divergent relationships with the constructs in the correlation matrix. The weighted visualization effectively complements the numeric coefficients by offering a more intuitive spatial representation of these relationships, helping us grasp both the general patterns and the subtleties in the data.
4. Personal Reflection😆
4.1 Limitations of the Study🥲
This study’s analysis reveals several challenges stemming from the interconnectedness of the six TAM constructs, which made it difficult to isolate the individual effects of demographic variables. The complex relationships between constructs like PU, PE, and PEU sometimes blurred the lines. Importantly, it became evident that multiple demographic factors could simultaneously influence students’ perceptions of translation technologies. As a result, the various plots—whether dot plots, bar plots, or line plots—along with the correlation matrix, yielded unexpected results that often diverged significantly from my initial assumptions. The hypotheses I set out with were not strongly validated by the data, which I consider the primary limitation of this study.
Most of the observations I made appear statistically negligible, suggesting that the impact of demographic factors on students’ technology acceptance may not be as prominent as initially anticipated. However, these findings still provide valuable insights, offering a new perspective on factors that could potentially influence students’ attitudes toward translation technologies. Although the statistical significance of these relationships may be limited, they highlight the need for further exploration into how demographic variables, like Family Residence and Gender, may subtly affect perceptions and adoption.
Additionally, my proficiency in RStudio, which I acknowledge as still developing, hindered my ability to perform a deeper analysis of the available data. Despite my efforts to utilize the software to its full potential, there were technical limitations that prevented me from uncovering a more detailed understanding of the data. Nonetheless, I remain confident that demographic variables do indeed play a role in students’ mastery of translation technologies.
Looking ahead, future research could address these limitations by diving deeper into the interaction between demographic factors and newer technological forms, especially as the landscape of translation technologies continues to evolve. Expanding the sample to include more diverse populations, or incorporating longitudinal studies, could offer richer insights into how shifts in family structures, education systems, or even cultural backgrounds influence students’ acceptance of these technologies.
4.2 Reflection on My Personal Learning and Growth🆙
This project has been an incredibly rewarding journey for me, one that has provided numerous opportunities for learning and personal growth. It was my first time managing a research project independently, and I found myself genuinely passionate about unlocking insights from the data and interpreting the results. While I encountered challenges along the way, these obstacles only fueled my desire to push further and refine my understanding of both the methods and the data.
Technical Skills: One of the most significant aspects of this project was the opportunity to enhance my technical skills, particularly in data analysis and visualization. I was able to integrate knowledge from the class, such as processing data into well-formatted tibbles and using ggplot2 to create a range of visualizations, including bar plots, dot plots, and line plots. These tools were essential for exploring and presenting the relationships within the data. I also gained hands-on experience in creating more advanced visualizations, such as relationship maps using nodes and edges, which allowed me to better understand how variables interact within a network.
In addition to applying these class-based techniques, I also learned new methods for visualizing and interpreting data. For example, I utilized correlation analysis and the Pearson model to assess relationships between variables. Interpreting the resulting graphs, especially the weighted relationship map, was challenging but immensely rewarding. It forced me to think critically about how to visually represent and convey the strengths and directions of these relationships. One key realization I had was that the sheer number of objects surveyed—four demographic variables and six TAM constructs—added a layer of complexity to the analysis that made interpretation much more difficult than anticipated. The dense interconnections between these variables posed significant challenges in isolating meaningful patterns.
Additionally, I also learned how to insert in-text citations and generate a reference list in a .bib file, which streamlined my process of citing sources and ensured proper referencing throughout my report. This skill not only made the process of writing more efficient but also helped me maintain academic rigor by properly crediting the works that informed my study.
Despite these challenges, I was able to navigate the complexities of the project with the help of various tools, including ChatGPT, which provided guidance when I encountered roadblocks in understanding or executing certain tasks. This support, combined with my determination, enabled me to successfully analyze and interpret the data, transforming raw numbers into meaningful insights. The entire process not only enhanced my technical abilities but also deepened my appreciation for the complexity of data analysis and visualization.
Critical Thinking and Problem-Solving: This study sharpened my ability to think critically and solve problems in a systematic way. I had to evaluate complex, interconnected relationships between demographic variables and TAM constructs, often revisiting and revising my approach. Analyzing seemingly simple demographic variables in relation to technology acceptance forced me to recognize the complexity of human behavior and technology interaction. For example, I initially hypothesized clear connections between certain demographic variables and TAM constructs, but the data proved far more complex than I had anticipated. This required me to adapt my thinking, consider alternative explanations, and refine my analysis methods to ensure that I was drawing accurate conclusions.
Emotional Journey: The emotional journey throughout this project was both rewarding and challenging. There were moments of frustration—particularly when interpreting results that diverged from my initial hypotheses or when I faced technical challenges with RStudio. However, these moments also provided an opportunity for growth, as I had to problem-solve and push through the frustration to find solutions. On the flip side, the sense of satisfaction when I completed the analysis and synthesized my findings was immensely rewarding. I felt a strong sense of accomplishment when I was able to present my results in a coherent way, knowing that I had worked through the difficulties and achieved something meaningful. This emotional rollercoaster reinforced my passion for research and motivated me to keep improving my skills and knowledge.
4.3 Acknowledgements💗
I am deeply grateful for the opportunity to take this course, which has significantly expanded my understanding of digital humanities and research methodologies. The guidance and support provided by the professor were invaluable, particularly in helping me navigate the complexities of data analysis, visualization, and developing a systematic framework in RStudio. I truly feel that this course has equipped me with the confidence and expertise to tackle future research challenges with a more nuanced and informed perspective. As Aristotle once said, “The more you know, the more you realize you don’t know,” and I am determined to continue honing my technological skills to keep pushing the boundaries of my knowledge.
5. References
Alcina, A. (2008). Translation technologies: Scope, tools and resources. Target. International Journal of Translation Studies, 20(1), 79–102. https://doi.org/10.1075/target.20.1.05alc
Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319–340. https://doi.org/10.2307/249008
Dianati, S., Taptamat, N., Uchiyama, A., & Akagawa, N. (2022). Factors that influence translation and interpreting technology adoption by university instructors: Through the lens of the technology acceptance model (TAM). Journal of Translation and Language Studies, 3(1), 12–28. https://sabapub.com/index.php/jtls/article/download/439/246
Li, X., Gao, Z., & Liao, H. (2024). An empirical investigation of college students’ acceptance of translation technologies. PLoS ONE, 19(2). https://doi.org/10.1371/journal.pone.0297297
Marangunić, N., & Granić, A. (2014). Technology acceptance model: A literature review from 1986 to 2013. Universal Access in the Information Society, 14(1), 81–95. https://doi.org/10.1007/s10209-014-0348-1
Salloum, S. A., Aljanada, R. A., Alfaisal, A. M., Al Saidat, M. R., & Alfaisal, R. (2024). Exploring the acceptance of ChatGPT for translation: An extended TAM model approach. In A. Al-Marzouqi, S. A. Salloum, M. Al-Saidat, A. Aburayya, & B. Gupta (Eds.), Artificial intelligence in education: The power and dangers of ChatGPT in the classroom (Vol. 144, pp. 527–542). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-52280-2_33
Yang, Y., & Wang, X. (2019). Modeling the intention to use machine translation for student translators: An extension of technology acceptance model. Computers & Education, 133, 116–126. https://doi.org/10.1016/j.compedu.2019.01.015
